Indeed. To nitpick a little and add some more information: the 25 bitcoins isn't necessarily the cost of electricity, it's the reward, and the expenses incurre

Indeed. To nitpick a little and add some more information: the 25 bitcoins isn't necessarily the cost of electricity, it's the reward, and the expenses incurred to reap that award approach 25 but likely sit slightly below it to allow for some profit, and the expenses aren't all electricity either but in large part also hardware and overhead. So what does the split look like in expenses (electricity, hardware, overhead etc). Well in the past few years hardware had a tremendous turnover rate as the industry essentially built an application specific integrated circuit for one particular hash, and then tried to catch up with decades of moore's law (I think they're now at 14nm), seeing extreme exponential growth of hashing power and some hardware going obsolete after a few months and not even being able to break even on their cost (i.e. the hardware:reward ratio was positive, which gives you a sense of how much was spent on hardware), meaning most money probably went to building new chips rather than burning electricity. Take January 2013 to January 2014 for example, hashing power increased from about 25k to 12.m, or 500x in 1 year. That meant if you were reaping 10% of the rewards in early 2013, a year later your setup could only reap 0.02% of the total rewards, necessitating tremendous investments to keep up with competition, investments that are obviously capped by bitcoin block reward, thus electricity was just a fraction of total expenditures. Of course this changes over time, they're now hitting some real walls and combined with a price stagnancy of bitcoin the hashing power. January 2014-2015 saw just a 21x increase in hashrate for example. So it's safe to say that today, a much greater share of the block reward goes to electricity than before. But even now, much is spent on replacing old equipment and on buying hardware that's just as powerful but uses less electricity than before, meaning even today not the entire 25 bitcoins are spent on electricity and thereby not the entire $7 either. Beyond that I think it's a bit misleading to state $7 per transaction, even though it's obviously correct in a certain sense that if you take the total expenditures and divide them by the units, you get an average. But if I run an automated subway between two parts of town for $100 per trip that leaves every 10 minutes regardless of occupancy, and it has a max capacity of say 1000 people, it'd be a bit misleading to say if one person took the subway and paid a $1 ticket that it costs $100 per person even though that's the average. No it costs $100 per subway trip whether it has 0 or 1000 people, and if 1 person gets on then it still costs just $1 per person. The fixed costs are $100, regardless of occupancy. Now bitcoin specifically does have a max 'occupancy' (transactions per block) which are close to $7, but bitcoin has 1mb blocks every 10 minutes. That's the bandwidth, 1.7 kb per seconds. We could have 100mb blocks and the 'electricity per transaction' would drop to 7c, while the bandwidth would still be just 170kb/s. As storage, bandwidth and cpu gets cheaper and bigger, we could eventually get to 1 gigabyte blocks and beyond. And as 1 transaction is just some data in a ledger, we could even use 1 transaction (i.e. some data) to represent pairs of multiple transactions. In short, while you're completely correct in citing the $7 figure, I think it's misleading to anyone who doesn't know anything about bitcoin because it tells this story of a ridiculously expensive and inefficient system, without mentioning a context where blocks are like a digital subway (referencing the above analogy) that we can make much bigger with the stroke of a keyboard (there are some genuine technical challenges but they're surmountable), a digital subway that's pretty much in the 56k modem stage and whose scaling up is long overdue, and thereby whose cost per transaction is way exaggerated. The biggest topic in bitcoin right now is essentially how to scale this thing up in a sensible way and once its agreed upon, that $7 will come down to a fraction of a penny before long. I've got a $60 TV+Internet package at home for example and my internet speed I just tested downloads at 10.5mb per second. Every 10 minutes bitcoin puts out one 1 mb block, in the same time I can download 6300. So even scaling by a factor of 100 right now on a residential line is feasible. There's a lot more to it of course, there are some political and possibly some security issues involved (mostly the more you scale it, the more only wealthy organisations can afford the connections necessary, the less ownership of the system by 'normal' people) but you'd have to really scale a lot before you get there. Simply scaling up by say 10x and then following normal bandwidth/storage growth rates a year would be a great start.